WOW! A great thread with many designers I respect. The seam I've been mining all my audio DIY life is phono stages.
There is so much reward for the effort and of course, a little frustration on the way :-)
As most here will know, a lot of analog's perceived "issues" are actually present in the phono stage and quite often attributed to the mechanical aspects as Thom alludes to.
The overload aspects of the phono stage encompasses all aspects of the design and is where many designers, not so here though, look at the basic requirements and assume the signal levels are low and limited to the audio stage
only. Compared to power amps they are but I think many look at headline specs, say a MM input of 5mV and leave it at
that. However take that figure at 1k and then project that to 20k which is 20dB higher and then things look way different. Include the ability to handle HF transients and things look different again. I only know any of this
through trying stuff out and listening and it is illuminating to have people put a technical perspective on this i.e. some of this stuff can be measured.
However the thing I design for is hard to measure OR may be measured if only we knew what to look at. I design for
the ability of the music to communicate it's message to me. Consequently I don't measure anything. Mainly because I can't as I don't have the equipment nor could I afford it of sufficient accuracy AFAIK. Knowing (I think) the
accuracy required to measure RIAA EQ deviation, I find it hard to believe some of the figures being claimed in this
thread. A deviation of 0.1dB is I think 1% accurate. To measure this accurately requires test instrumentation to be
an order of magnitude more accurate so that's 0.1%. Is test equipment of this caliber being used to verify this? And
regularly checked against a test standard? As an amateur, I go for the "model it" approach aiming for the best fit
to the curve and then used better than 1% parts to hopefully get within the 1% window so within the 0.1dB deviation which could be called +/- 0.05dB. Is it? I have no idea but as I have got the model better and my design has got better, it would seem so.
Talking abou the EQ curve, there is the matter of the cutterhead rolloff or the 3.18uS turnover point. I used to
approach this as a lone point however as someone pointed out to me once, this turnover point affects the curve
waaaay lower down and so now I sim it as a LP filter that my inverse EQ feeds the sim with. Correcting for it's
effects lower down got me closer to a better phono stage. So I would question in some way, the effects of making it
switchable IF the other components in the EQ are not also altered at the same time.
I found it real interesting Johnothan's calcs re the effect of cart resonance and the frequencies it occurs at. It got
me thinking about its effects on not only the audio cct but the power supply. I personally don't like regulation (we
all have our prejudices) as I like to think the power supply should be as Nic Doshi says, as benign as possible. My experience is that this is really hard and when using regulation, for me at least, it has proven to be impossible.
With all this HF stuff going on, the concept that most regulators have a vice like grip on proceedings is a myth and their contribution to the circuit becomes nearly as great as the audio circuit itself. I try and avoid things like that ... well in my head anyway :-)
It will be no surpirse therefore that I favour hollow state. Another area that seems to be a poor cousin in the
power supply design is the heaters. IME, they have as much inluence over the sonics as the HT and so in mine, they
get as much filtering as the HT. It's not surprising that many commercial offerings kinda pass this by as it becomes
real expensive to do. I have no idea what anyone here does so it's not a criticism, just an observation of those
circuits I've seen.
I find it at odds with my experience that you can have a great design that makes good records sound good and also
bad resordings sound good. IME, a great design makes all records sound better and that's not done by introducing
"flavour" by being coloured but by making it more technically able. Bad recordings, to me anyhow, usually present more of a challenge and a better design is able to meet this challenge without causing the circuit to hold a white flag up and sound horrible. As the design gets better, good recordings sound even better, bad recordings can also benefit and rise above awful to enjoyable BUT the real essence of a good design is that all recordings sound more different. It is this aspect that drives me on. The better the design, the more of my record collection opens up to become enjoyed.
Some that used to be shocking I have discovered are real treats now. This is why I focus on the communication of the
music being the sole arbiter of goodness in the design and as such, the rest seems to come with it.
Another thing I wonder about is the concept of channel seperation. A cartridge is only so so in this regard. Once I used to do dual mono re the PSU but now I use a single supply as this seems to ground the musicians better and they seem to play together better. Its seems to be one of those HiFi vs music trade offs. I also wonder how you can get two supplies to be perfectly the same regarding noise and grounding and so therefore be exactly the same at all frequencies. As I can't see how to do this and the results of a single supply in my designs to be superiour, I wonder how much we should chase this notion of chqannel seperation.
Thanks Thom for kicking this thread off. I hope it reveals a bit more as it goes. I just wish I came to find it earlier.
regards,
Stephen
www.izzy-wizzy.com/audio
There is so much reward for the effort and of course, a little frustration on the way :-)
As most here will know, a lot of analog's perceived "issues" are actually present in the phono stage and quite often attributed to the mechanical aspects as Thom alludes to.
The overload aspects of the phono stage encompasses all aspects of the design and is where many designers, not so here though, look at the basic requirements and assume the signal levels are low and limited to the audio stage
only. Compared to power amps they are but I think many look at headline specs, say a MM input of 5mV and leave it at
that. However take that figure at 1k and then project that to 20k which is 20dB higher and then things look way different. Include the ability to handle HF transients and things look different again. I only know any of this
through trying stuff out and listening and it is illuminating to have people put a technical perspective on this i.e. some of this stuff can be measured.
However the thing I design for is hard to measure OR may be measured if only we knew what to look at. I design for
the ability of the music to communicate it's message to me. Consequently I don't measure anything. Mainly because I can't as I don't have the equipment nor could I afford it of sufficient accuracy AFAIK. Knowing (I think) the
accuracy required to measure RIAA EQ deviation, I find it hard to believe some of the figures being claimed in this
thread. A deviation of 0.1dB is I think 1% accurate. To measure this accurately requires test instrumentation to be
an order of magnitude more accurate so that's 0.1%. Is test equipment of this caliber being used to verify this? And
regularly checked against a test standard? As an amateur, I go for the "model it" approach aiming for the best fit
to the curve and then used better than 1% parts to hopefully get within the 1% window so within the 0.1dB deviation which could be called +/- 0.05dB. Is it? I have no idea but as I have got the model better and my design has got better, it would seem so.
Talking abou the EQ curve, there is the matter of the cutterhead rolloff or the 3.18uS turnover point. I used to
approach this as a lone point however as someone pointed out to me once, this turnover point affects the curve
waaaay lower down and so now I sim it as a LP filter that my inverse EQ feeds the sim with. Correcting for it's
effects lower down got me closer to a better phono stage. So I would question in some way, the effects of making it
switchable IF the other components in the EQ are not also altered at the same time.
I found it real interesting Johnothan's calcs re the effect of cart resonance and the frequencies it occurs at. It got
me thinking about its effects on not only the audio cct but the power supply. I personally don't like regulation (we
all have our prejudices) as I like to think the power supply should be as Nic Doshi says, as benign as possible. My experience is that this is really hard and when using regulation, for me at least, it has proven to be impossible.
With all this HF stuff going on, the concept that most regulators have a vice like grip on proceedings is a myth and their contribution to the circuit becomes nearly as great as the audio circuit itself. I try and avoid things like that ... well in my head anyway :-)
It will be no surpirse therefore that I favour hollow state. Another area that seems to be a poor cousin in the
power supply design is the heaters. IME, they have as much inluence over the sonics as the HT and so in mine, they
get as much filtering as the HT. It's not surprising that many commercial offerings kinda pass this by as it becomes
real expensive to do. I have no idea what anyone here does so it's not a criticism, just an observation of those
circuits I've seen.
I find it at odds with my experience that you can have a great design that makes good records sound good and also
bad resordings sound good. IME, a great design makes all records sound better and that's not done by introducing
"flavour" by being coloured but by making it more technically able. Bad recordings, to me anyhow, usually present more of a challenge and a better design is able to meet this challenge without causing the circuit to hold a white flag up and sound horrible. As the design gets better, good recordings sound even better, bad recordings can also benefit and rise above awful to enjoyable BUT the real essence of a good design is that all recordings sound more different. It is this aspect that drives me on. The better the design, the more of my record collection opens up to become enjoyed.
Some that used to be shocking I have discovered are real treats now. This is why I focus on the communication of the
music being the sole arbiter of goodness in the design and as such, the rest seems to come with it.
Another thing I wonder about is the concept of channel seperation. A cartridge is only so so in this regard. Once I used to do dual mono re the PSU but now I use a single supply as this seems to ground the musicians better and they seem to play together better. Its seems to be one of those HiFi vs music trade offs. I also wonder how you can get two supplies to be perfectly the same regarding noise and grounding and so therefore be exactly the same at all frequencies. As I can't see how to do this and the results of a single supply in my designs to be superiour, I wonder how much we should chase this notion of chqannel seperation.
Thanks Thom for kicking this thread off. I hope it reveals a bit more as it goes. I just wish I came to find it earlier.
regards,
Stephen
www.izzy-wizzy.com/audio